Help protect the Great Barrier Reef with TensorFlow on Kaggle Join Challenge

quality

  • Description:

QuALITY, a multiple-choice, long-reading comprehension dataset.

We provide only the raw version.

Split Examples
'dev' 230
'test' 232
'train' 300
  • Features:
FeaturesDict({
    'article': Text(shape=(), dtype=tf.string),
    'article_id': Text(shape=(), dtype=tf.string),
    'difficults': Sequence(tf.bool),
    'gold_labels': Sequence(tf.int32),
    'options': Sequence(Sequence(Text(shape=(), dtype=tf.string))),
    'question_ids': Sequence(Text(shape=(), dtype=tf.string)),
    'questions': Sequence(Text(shape=(), dtype=tf.string)),
    'set_unique_id': Text(shape=(), dtype=tf.string),
    'source': Text(shape=(), dtype=tf.string),
    'title': Text(shape=(), dtype=tf.string),
    'topic': Text(shape=(), dtype=tf.string),
    'url': Text(shape=(), dtype=tf.string),
    'writer_id': Text(shape=(), dtype=tf.string),
    'writer_labels': Sequence(tf.int32),
})
@article{pang2021quality,
  title={ {QuALITY}: Question Answering with Long Input Texts, Yes!},
  author={Pang, Richard Yuanzhe and Parrish, Alicia and Joshi, Nitish and Nangia, Nikita and Phang, Jason and Chen, Angelica and Padmakumar, Vishakh and Ma, Johnny and Thompson, Jana and He, He and Bowman, Samuel R.},
  journal={arXiv preprint arXiv:2112.08608},
  year={2021}
}

quality/raw (default config)

  • Config description: Raw with HTML.

  • Dataset size: 22.18 MiB

  • Examples (tfds.as_dataframe):

quality/stripped

  • Config description: Stripped of HTML.

  • Dataset size: 20.73 MiB

  • Examples (tfds.as_dataframe):